Viewing profile — milkshakes

you can't downvote a submission it's just that nobody upvoted it

> models are still relatively easy to monitor and constrain are they?

they also have a pretty handy button right next to the error to ftfy

i guess what i'm curious about is where you would draw the line, and why. how do you define a user agent? is it desirable or not for users to be able to discover and access resourc…

i guess the thing i'm most confused about is what is the higher level goal here. 1 in 8 humans on the planet are experiencing the "web" through chatgpt alone. many have migrated to…

start, not finish

http://www.incompleteideas.net/IncIdeas/BitterLesson.html > One thing that should be learned from the bitter lesson is the great power of general purpose methods, of methods that c…

start here: https://huggingface.co/blog/mlabonne/abliteration

what top comment?

nyt is no fan of AI, but they cover this: https://www.nytimes.com/2026/07/09/business/china-russia-ai-...

this is quite literally reward hacking. the model, under evaluation with cyber capabilities enabled, used those capabilities to simply bypass the exercise entirely and aim straight…

who are you talking about? again, my question is concerned with the "second class labs" and the sustainability of the distillation-as-a-service model.

it is not a required first step for training a model, sure. but that's not what i claimed. what i claimed is that is how they are so significantly _reducing the cost_ of training o…

the obvious difference is the massive scale of data and compute required to develop and evolve these models, and the costs they impose on those building them.

https://www.anthropic.com/news/detecting-and-preventing-dist... Moonshot AI Scale: Over 3.4 million exchanges The operation targeted: Agentic reasoning and tool use Coding and data…

i never assumed that, and i do keep up with the publications. i'm also not saying it's a dumb thing to do! what i am saying is that empirically, it appears that distillation of a m…

the question was: what is the endgame for the stated "second class labs" strategy of distilling their frontier competitors then undercutting them on price?

the USG/NSA will fund chinese labs? to what end?

> with budgets and what will fund these budgets exactly? inference is cheap, distillation is cheap, training is what's expensive.

assume you are a "second class lab" and you are in fact making progress by distilling the results of the frontier labs' efforts. what is the end game for this strategy? if the fron…

is this a joke?

in fact, telegram does support e2e encryption ("secret chats")